Defects and electrical properties in Bi-doped calcium manganite

The air synthesized samples Ca1–xBixMnO3–δ, where x = 0.05, 0.10 and 0.15, show the formation of fully oxidized solid solution, δ = 0, with the orthorhombic structure (S G Pnma). It is shown that electrical properties at temperatures below 700 K are mostly governed by the charge disproportionation and local Jahn–Teller distortions of octahedral oxygen environment of Mn3+ cations. The bismuth doping strongly affects the disproportionation enthalpy, charge carriers concentration and conductivity. The metallic-like temperature dependencies of electrical conductivity and thermopower are interpreted within frameworks of a small polaron mechanism of electron transport. The maximum power factor 250 μW·K–2·m–1 is observed in Ca0.85Bi0.15MnO3–δ at 1120 K.


Introduction
Oxide thermoelectric materials are chemically and thermally stable at elevated temperatures in oxidative atmospheres and, therefore, very attractive for thermoelectric utilization of high-temperature waste heat. They are also cost efficient and nontoxic compared to the state-of-the-art chalcogenides, silicon germanium compounds, skutterudites, Heusler alloys, etc [1,2]. In this connection and owing to inherently large thermopower perovskitetype calcium manganite CaMnO 3 has been identified as a promising material for the development of n-type thermoelectrics [3][4][5]. Considering rather low electrical conductivity a lot of efforts have been put into studies of different doping strategies aimed at conductivity enhancement and increase of the overall thermoelectric efficiency of CaMnO 3 . The Ca-site substitutions by Y, La, In, Sn, Sb, Pb, Bi and rare-earth metals have been examined in works [3,[6][7][8]. The data on replacement of Mn by Mo, Ru, Nb and Ta are presented by authors [9][10][11]. Notice that the concentration dependent maxima of conductivity and thermoelectric efficiency do not coincide as a rule. This takes place because simultaneously with changes in the amount of charge carries doping cations influence the overlap of Mn3d and O2p orbitals and bending of O-Mn-O chemical bonds and, thus, impact the width of the conducting band [5,8,12,13]. Another doping effect in case of replacement of manganese is related with disruption of conducting -Mn-O-Mnchains in the structure, which usually results in decline of the mobility of electronic carriers [11,14]. Though both types of doping may improve conductivity, the doping of Ca sites appears to be preferable.
Among numerous dopants bismuth, which is much heavier compared to calcium, attracts particular interest. The large mass difference is favorable for excitation of local phonon modes near doping centers, strong phonon scattering and reduction of thermal diffusivity of Ca 1−x Bi x MnO 3 . Several works are known where hightemperature conducting and thermoelectric properties of solid solutions Ca 1-x Bi x MnO 3 have been studied [15][16][17][18]. However, significant disagreements are observed as to which composition is the best thermoelectric. For instance, authors [16] propose x=0.03 as having the minimum thermal conductivity while a better properties combination in work [15] is found at x=0.04. The divergence may possibly be related with differences in defect state and heterostructures at nanoscale [19,20]. In this regard we can notice also that the parent manganite and its doped derivatives, though structurally robust, rather actively participate in oxygen exchange with the ambient atmosphere at elevated temperatures, and this process may considerably influence structural features and transport characteristics [21,22]. Correspondingly, the formula of the solid solution may be more correctly presented as Ca 1−x Bi x MnO 3-δ , where symbol δ stands for oxygen non-stoichiometry.
In order to gain a deeper understanding of how temperature driven defect formation reactions, variations of bismuth content and oxygen non-stoichiometry may contribute to changes in concentration and mobility of electronic carriers we performed measurements of electrical conductivity and thermopower in Ca 1-x Bi x MnO 3-δ at heating in air to 1250 K. For the data analysis we applied the earlier developed model of high-temperature polaron transport in CaMnO 3-δ [23].

Methods
The samples in oxide series Ca 1-x Bi x MnO 3-δ , where x=0.05, 0.10 and 0.15, were obtained via organo-metallic precursors. For the synthesis, the starting powder reagents CaCO 3 (99.8%), Bi 2 O 3 (99.8%) and Mn 2 O 3 (99.9%) were weighed in desirable proportions, placed in a quartz beaker and dissolved in nitric acid, whereupon glycine С 2 H 5 NO 2 was added to the solution in 50 mol% excess to the total amount of metal cations. Gradual evaporation on a hot plate resulted in ignition and smooth burning of the desiccated gel. The resulting residue was carefully grinded with a mortar and pestle and calcined at 1173 Kfor 10 h in air for removal of trace organics and carbon. The resulting powder material was milled again and pelletized under uniaxial load of 200-300 MPa followed by sintering at 1573 K for 10 h in air. The pressed pellets were covered with the powdered material and placed in a crucible with the lid in order to avoid any changes of chemical composition at this stage of hightemperature treatment. The density of prepared ceramics was calculated from the geometric dimensions and mass of polished samples.
Powder x-ray diffraction (XRD) in Bragg-Brentano mode was utilized for phase purity control and structure analysis. XRD data were collected with the help of a XRD-7000 (Shimadzu) diffractometer (CuK α radiation) in the 2Θ range 20°-80°with 0.03°step size and 3 s acquisition time. The calculations of the elementary cell parameters were performed using PCW 2.4 calculation package [24].
The scanning electron microscopy (SEM) with a secondary electron detector JEOL JSM 6390LA was used for the analysis of morphology and homogeneity of the sintered samples.
A Setaram TG-92 thermoanalyzer was used in order to collect weight changes of the samples equilibrated at 1223 K for half an hour and cooled down with the rate of 1 K min −1 in air. This mode of thermal treatment results in fully oxidized samples with δ=0 at room temperature [5,25] while the isobar of cooling is believed to coincide with temperature dependent variations of equilibrium oxygen content in the sample. Therefore, the changes of oxygen non-stoichiometry with temperature can be calculated as where Δm(T) is the weight change relative to the state δ=0 of the initial sample with the weight of m s at room temperature, and Ms and M O stand for the molar weight of the sample and oxygen, respectively. Typically, the weight m s was about 100 mg, and the uncertainties in determination of δ did not exceed 0.003. Rectangular bars 2×2×12 mm 3 were cut from the sintered pellets for electrical measurements from 300 to 1250 K in air. One specimen, equipped with butt S-type thermocouples, was used for measurements of thermopower. The temperature gradient in the furnace along the sample was about 15 K cm −1 . The Pt leads of the thermocouples were used as voltage probes. Thermopower data were corrected for the contribution of platinum [26]. Another specimen was used in four-probe measurements of d.c. conductivity. Current leads of Pt wire (0.3 mm) were tightly wound to the sample at 10 mm spacing while the spacing between the potential probes was 8 mm. The electrical parameters were measured with a high-precision Solartron 7081 voltmeter. More experimental details can be found elsewhere [27]. The uncertainties in the obtained results are estimated to be below 5%. The collected data were nalysed with the help of SigmaPlot V12.5 media [28].

Results and discussion
XRD patterns for the air synthesized samples Ca 1-x Bi x MnO 3-δ , where x=0.05, 0.10 and 0.15, show formation of single-phase perovskite-type oxides with orthorhombic structure (S G Pnma), figure 1. The replacement of Ca 2+ (R CN12 =1.340 Å) for larger Bi 3+ (R CN12 =1.450 Å) and charge compensating appearance of Mn 3+ (R CN6 =0.645 Å) in place of smaller Mn 4+ (R CN6 =0.530 Å) cations is accompanied with the increase of the crystalline structure parameters, table 1, in a good agreement with the available data [15,16]. The representative SEM micrographs of the sintered materials x=0.10 and x=0.15 in figure 2 reveal pores in the sintered samples. The measured density of the samples is about 85%-88% of theoretical value, table 1. More important for reliable electrical measurements is to observe well sintered grains in figure 2 and a uniform distribution of the constituent elements as shown in figure 3.
The temperature dependent variations of oxygen content (3-δ) in Ca 1-x Bi x MnO 3-δ are shown in figure 4 as calculated from TG measurements in air. It is seen that the oxygen depletion takes place at temperatures above 700 K, and the increase in bismuth content is followed with the decrease in the amount of oxygen released in air. The respective reaction can be represented as where Kröger-Vink notations are used for defect species [29]. The bismuth doping results in the formation of electronic defects ¢ Mn Mn (Mn 3+ cations) and, therefore, is favorable for the shift of the equilibrium (2) to the left hand side, i.e. for the decrease of the oxygen loss in accord with the experimental data in figure 4.
The measured data for electric conductivity σ and thermopower S in figure 5 agree with the literature data [15][16][17][18]. The porous morphology of the samples may result in about 15%-20% smaller values of the measured conductivity compared to perfectly dense ceramics [30]. At the same time, the porosity does not greatly affect the   conductivity activation energy, thermopower and other important characteristics. The appearance of bismuth in the crystalline structure is accompanied with a considerable increase of the conductivity from about 1 S cm −1 at x=0 [5,31] to 300 S cm −1 at x=0.15 at near room temperature. The conductivity variations with temperature exhibit metal-like behavior in a wide temperature range from about 250 to 1100 K. On the other hand, the conductivity values are quite small even in comparison with poor metals, and at the same time,  absolute values of thermopower are much larger than in metals. Notice additionally that the orthorhombic structure persists in the parent manganite at heating to about 1170 K, while donor dopants shift the ortho↔tetra transition to even higher temperatures [5,21]. Therefore, the data in figure 5 reflect mainly temperature dependent changes of conductivity and thermopower in the orthorhombic structure. Only above 1150 K one can observe small upturns of the thermopower plots that can signal incipient transition to the tetragonal structure. The electric properties combination observed in the experiment can be explained within frameworks of a polaron conduction mechanism in manganites [32][33][34]. Depending on the polarizability of the material, the charge carriers may be associated with either 'large' or 'small' polarons [35]. Authors [36] argue that electron-doped manganite compounds are near a large−to small−polaron crossover. At elevated temperatures, where the effects of dynamic disordering and localization [37] are especially pronounced, the picture of electron transport caused by the movement of small polarons seems preferable [22,38]. In this relation we have to notice that apparently metallic transport characteristics can be observed in small-polaron systems with the activation energy of about k B T so that the pre-exponential may overbalance in the expression for polaronic conductivity [39] where s 0 is a coefficient, k B is the Boltzmann constant, and s E is the conductivity activation energy. Similarly to the other donor-doped manganites with s E =10-40 meV [3,5,16], the function log(σT) versus 1/T is increasing at heating to 700 K, figure 6, i.e. in conditions where δ in Ca 1-x Bi x MnO 3-δ is near zero, figure 4. More expressed increase of the plots at further heating reflects commencement and intensification of oxygen exchange with ambient atmosphere, the increase of the formation of electronic defects in reaction (2) and respective enhancement of conductivity. Notice also that variations of δ and, consequently, of electron concentration at T=const in figure 4 become smaller with the increase of bismuth content. In the result, the upward bends of the plots in figure 6 are less pronounced at x=0.10 and 0.15 compared to x=0.05. The negative thermopower S is in accord with n-type conductivity in Ca 1-x Bi x MnO 3-δ , figure 5(b). The thermopower changes with dopant concentration and temperature can be interpreted with the help of a general expression [39] for hopping conductivity where e is the elementary charge, g and n represent the amount of positions available for jumps and concentration of n-type charge carriers, respectively, and E S denotes the thermopower activation energy. First of all, it is seen from (4) that larger bismuth content, i.e. larger n's, must favor smaller |S| values as, indeed, is observed in the experiment, figure 5(b). The compositional changes due to oxygen loss (2) above 700 K also result in a decrease of the thermopower absolute values. At lower temperatures, where oxygen content is permanent, variations of thermopower may reflect changes caused by the charge disproportionation reaction Supplementing these relations with the equilibrium constant for reaction (5) we obtain a system of equations that can be resolved as  (5), respectively, and R is the gas constant. It follows from (9) and (10) that a decrease in the concentration n of Mn 3+ cations with the increase of temperature may take place only whenDH D 0 >0, which is consistent with the endothermal character of reaction (5). For quantifying calculations we have to notice that polaron jumps from Mn 3+ to Mn 4+ cations are energetically favorable on condition of equal multiplicities of the initial Mn 3+ and final Mn 4+ spin configurations [12]. Therefore, Mn 4+ cations that do not satisfy this requirement are not available for the jumps, and must be considered as excluded from the transport process while equation (4) ought to be modified as Unfortunately, direct computations of the concentration g ex of the excluded sites are difficult because exact statistics of spin distribution over manganese cations must be known. However, the backward estimation can be made by fitting experimental data in figure 5(b) with the help of (12) and g and n taken from (9). For simplicity, fitting parameters g ex , E s , DH D 0 and DS D 0 can be assumed temperature independent. The trial attempts to treat the data in figure 5(b) below 700 K showed that E s can be safely set to zero. Notice here that the observation of the thermopower activation energy smaller compared to the conductivity activation energy gives additional confirmation to the small polaron character of charge transport in the manganites [39]. The obtained values of g ex , DH D 0 and DS D 0 in table 2 make it possible to describe with a good precision the experimental polytherms for thermopower at different bismuth content in Ca 1−x Bi x MnO 3 , figure 5(b).
The enthalpy change DH , D 0 which is the parameter that most strongly affects temperature variations of thermopower, is increasing with bismuth content, table 2. The similar trend was observed earlier for the charge disproportionation reaction in Ca 1-x Pr x MnO 3 [23]. In perovskite-type manganites this value is approximately equal to the energy splitting Δε of manganese e d : 3 g z 2 and e d : 3 g y x 2 2states. The component e d : 3 g z 2 is lower on energy scale, and Mn 3+ O 6 octahedra are elongated along zaxis due to the Jahn-Teller effect [12]. The donor doping of CaMnO 3 is accompanied with partial transformation of highly symmetrical Mn 4+ O 6 to less symmetrical and larger Mn 3+ O 6 octahedra. In turn, the appearance of large and stretched Mn 3+ O 6 octahedra results in the overall expansion of crystalline structure and local deformations of Mn 4+ O 6 octahedra so that the gap between e d : 3 g z 2 and e d : 3 g y x 2 2states tends to increase with the doping. Notice also that at equal dopant content the enthalpy DH D 0 for Ca 1-x Bi x MnO 3 in table 2 is larger compared to Ca 1-x Pr x MnO 3 [23]. It is because larger size of Bi 3+ results in larger local deformations in the crystalline structure.
The thermodynamic parameters DH D 0 and DS D 0 in table 2 can be used further in order to calculate the equilibrium constant ( ) K T D according to (10) and find the temperature driven variations in the concentration of different manganese species from (9). The respective plots for Mn 3+ and Mn 4+ are shown in figure 7 where one can see that the disproportionation reaction (5) noticeably affects relative concentrations of manganese states above room temperature. Accordingly, at lower temperatures where n=x, the concentration of Mn 3+ cations depends only on the concentration of bismuth donors in Ca 1-x Bi x MnO 3 [23,43]. At the same time, the heating is favorable for a slight increase of [Mn 4+ ] av =g-g ex and a decrease of n so that these simultaneous changes result in the apparently <<metallic>> increase of the absolute values of thermopower in figure 5(b) up to the temperatures of oxygen depletion.
The temperature dependent changes in the concentration of mobile charge carries n and positionsg g ex available for polaron jumps in figure 7 can be applied for calculations of the coefficient σ 0 in equation (3) and where r is the length of polaron jumps, n 0 is the characteristic frequency of jump attempts, and N is the amount of formula units per cubic centimeter. Neglecting weak temperature dependence of r andn 0 we can represent equation (3) in the form more convenient for calculations and m E is the mobility activation energy. Parameters s and m E can be derived from the plots ( ·( )) / s -T n g g log ex versus 1/T, and the respective values in table 2 result in a quite satisfactory coincidence of the calculated and experimental plots of the conductivity in figure 5(a) up to the temperatures of oxygen take-off.
The obtained multiplier  s can be utilized for the estimation of the frequency n = 0 3.8·10 14 s -1 of polaronic jumps, which is consistent with the earlier data for Ca 1-x Pr x MnO 3 [31] and characteristic frequency ∼10 14 s -1 of optical phonons in similar perovskites [44]. Therefore, the small polaron model gives correct and consistent description of electron transport in Ca 1-x Bi x MnO 3 at temperatures up to at least 700 K. The temperature dependent plots for the mobility of n-type charge carries are shown in figure 8 as obtained from the experimental results for the conductivity in figure 5(a) and calculated data for n, i.e. the concentration of Mn 3+ cations, in figure 7. The mobility values below the characteristic threshold ∼1 cm 2 V -1 s -1 [3] separating narrow and broad band conduction types corroborate polaronic conductivity in Ca 1-x Bi x MnO 3 . The maxima in figure 8 take place because the mobility exponent in (14) is increasing with heating below ∼400 K. Further decline of the mobility with heating is due to the decrease of the pre-exponent ∼( Generally, the mobility plots are similar to those for Ca 1-x Pr x MnO 3 [31].  The power factor s = PF S 2 can be obtained from the measured data for thermopower and conductivity, figure 9. In difference with x=0.05 and 0.10, where PF varies rather weakly even at temperatures of intensive oxygen exchange, the power factor values in x=0.15 appreciably increase with temperature showing no anomalies at heating above 700 K. This difference in temperature dependent behavior may possibly reflect enhancement of local deformations of Mn 3+ O 6 octahedra and respectively induced increase of DH D 0 at deep doping. Consequently, the power factor in x=0.15 attains about 250μW·K -2 ·m -1 at 1120 K. This value is not record high yet as shown in figure 10. However, efforts towards improved microstructure and sintering may result in further enhancement of thermoelectric properties of the bismuth doped manganites.

Conclusion
The samples Ca 1-x Bi x MnO 3 , where x=0.05, 0.10 and 0.15, with the orthorhombic structure (S G Pnma) are obtained via organo-metallic precursors. The measurements of oxygen content, electrical conductivity and thermopower are carried within 298-1223 K in air. The oxygen take-off temperature at heating in air is found to increase with bismuth content. It is argued that metallic-like temperature dependencies of the conductivity and thermopower can be interpreted within frameworks of a small polaron model of electron transport in Ca 1-x Bi x MnO 3 . The bismuth doping is favorable for the increase of the charge carriers concentration and conductivity. At the same time, the conductivity tends to decrease with the temperature increase due to intensification of charge disproportionation of Mn 3+ cations. The disproportionation enthalpy is shown to depend rather strongly on bismuth content. As a result, the power factor in Ca 0.85 Bi 0.15 MnO 3-δ is a function noticeably increasing with temperature and attaining the largest value of 250 μW·K -2 ·m -1 at 1120 K.